HIV Infection, Vascular Disease, and Stroke

Abstract

With the transformation of HIV infection into a chronic disease for the majority of patients with access to combination antiretroviral therapy, HIV-infected individuals and their medical providers are facing a new set of challenges, many of which are more typical of an aging population. Accumulating evidence indicates that rates of stroke, like coronary heart disease, may be higher in HIV-infected individuals. This review will discuss the epidemiology of stroke in HIV infection, potential mechanisms underlying the pathogenesis of stroke, vascular risk and cognitive impairment, cardiovascular and stroke risk prediction, and strategies for stroke prevention in this at-risk population, with a focus on HIV-associated cerebrovascular disease in the current era of highly effective combination antiretroviral therapy

Highly effective combination antiretroviral therapy (cART) has ushered in an era of the HIV epidemic in which chronic infection is achievable for the majority of adherent individuals. Among those with well-controlled HIV infection, mortality rates are comparable to the general population.^1,2 As life expectancy increases,³ the focus of HIV care has shifted to include non-AIDS-defining illnesses, many of which are more typical of an aging population.^4,5 Cardiovascular disease (CVD) event rates, for example, are increased in HIV infection,^6–11 and mounting evidence indicates that a similar association exists between HIV and cerebrovascular disease (CVD). Progress in HIV-associated stroke research, however, has been hampered by various factors inherent to CVD, including its heterogeneous nature. Although ischemic stroke and intracerebral hemorrhage (ICH) have distinct risk profiles and pathogenetic mechanisms, and thus the effect of HIV on the risk of each outcome may differ, a composite endpoint of CVD^12,13 or even CVD combined with cardiovascular events¹⁴ is commonly used in studies of HIV-infected individuals. As the HIV-infected population survives longer and reaches advanced age, the clinical and public health implications of an elevated risk of CVD are expected to grow in relevance. The conflation of an aging HIV-infected population, the growing burden of CVD in resource-limited settings, where stroke incidence and associated mortality and disability exceed that in developed nations,^15,16 and the emergence of non-AIDS-defining illnesses^17,18 may pose a formidable challenge in caring for HIV-infected individuals worldwide.

The Epidemiology of Stroke in HIV Infection

Early estimates of stroke prevalence in HIV infection were based on studies with potential limitations.^19–23 Some were confounded by advanced immunosuppression and associated opportunistic infections and malignancy, others originated from nonpopulation based studies, and still others relied on autopsy data from AIDS patients without clinically apparent CVD or used flawed comparator groups. Of these early estimates, several supported an increased risk of stroke in HIV infection,^20,24,25 while others did not reach the same conclusion.^26–28 One of the few population-based studies predating cART evaluated data from the Baltimore-Washington Cooperative Young Stroke registry.²⁵ This study demonstrated an annual incidence of ischemic stroke and of ICH among AIDS patients of 0.2% each, although these rates were based on few events (n = 12). After excluding five cases with concomitant endocarditis, opportunistic infections or malignancy, the annual incidence of ischemic stroke was 0.14% and of ICH was 0.11%, and the adjusted relative risk (RR) for ischemic stroke was 9.1 (95% confidence interval [CI] 3.4–24.6) and for ICH 12.7 (95% CI 4.0–40.0). This early study specifically addressed the question of stroke risk in AIDS, as HIV-infected individuals without AIDS were included in the control group.

Evidence from the current era of highly effective cART more strongly supports an association between HIV and stroke (Table 1). Analysis of discharge data from the U.S. Nationwide Inpatient Sample found that the proportion of ischemic strokes that occurred in HIV-infected individuals increased by more than twofold from 1997 to 2006, adjusting for the rise in the absolute number of HIV-infected individuals during that time.²⁹ This finding, however, may have been confounded by the secular trend of improved survival as cART was introduced. In a large population-based Danish cohort, the adjusted incidence rate ratio (IRR) for a composite outcome of cerebrovascular events (i.e., cerebral infarction, unspecified stroke, transient ischemic attacks, subarachnoid hemorrhage, and ICH) among HIV-infected intravenous drug users (IVDU) was 3.94 (95% CI 2.16–7.16) and 1.60 (95% CI 1.30–1.95) for HIV-infected non-IVDU.³⁰ Smoking data, however, were not available in this cohort. In the Partners HIV cohort, derived from a clinical care database of two urban U.S. hospitals, HIV-infected individuals experienced higher rates of ischemic stroke (5.27 per 1,000 person-years [PY]) compared with uninfected controls (IRR of 1.40, 95% CI 1.17–1.69).³¹ The increased hazard in the HIV cohort was attenuated after adjustment for demographics and recognized stroke risk factors, including smoking, but remained significantly higher compared with uninfected controls (hazard ratio [HR] 1.21, 95% CI 1.01–1.46). Age-stratified analyses demonstrated a relatively greater hazard of ischemic stroke among younger HIV-infected individuals. In contrast, the risk of stroke in the Swiss HIV Cohort increased with advancing age.⁴ The hazard of stroke for individuals 65 years of age and above was almost 18 times greater than for those younger than 50 years, and the hazard for individuals between 50 and 64 years was 4 times greater. Despite important limitations present in each of these epidemiological studies, including incomplete data on relevant stroke risk factors and reliance on International Classification of Diseases (ICD) codes to identify cerebrovascular events, these findings support an elevated risk of ischemic stroke in HIV infection beyond that explained by traditional stroke risk factors. Ultimately, adequately powered population-based studies of stroke incidence using validated outcomes will be essential to advance our understanding of the burden of stroke among HIV-infected individuals.

Table 1.

Incidence rates of cerebrovascular events in HIV-infected individuals from recent epidemiological studies (2008–2013)

Author	Study population	Design	Number of HIV-infected individuals	Controls	Incidence rate or cumulative incidence (95% CI)	Adjusted rate ratio (HIV compared with uninfected controls) (95% CI)
Cerebrovascular eventsa
Bedimo et al, 201187	U.S. Veterans Health clinical case registry	Clinical care cohort	19,424	NA	11.68 (10.93–12.48) per 1000 PY	NA
Rasmussen et al, 201130	Danish HIV cohort	Population-based cohort	5031	45,279	Non-IVDU 10-year cumulative incidence 2.68 (2.16–3.30) IVDU 10-year cumulative incidence 3.08 (1.76–4.98)	Non-IVDU 1.60 (1.30–1.95) IVDU 3.94 (2.16–7.16)
Mateen et al, 2013144	Multicenter AIDS Cohort	Prospective cohort	1862	2169	3.3 per 1000 PY	3.3 per 1000 PY in HIV-infected versus 1.7 per 1000 PY in uninfected controls
Sabin et al, 201313	D:A:D	Prospective cohort	33,301	NA	1.4 per 1000 PY	NA
Vinikoor et al,201340	UNC HIV clinical cohort	Clinical care cohort	2515	NA	3.87 (2.90–5.06) per 1000 PY	NA
Ischemic stroke
Corral et al, 200939	Spanish HIV clinic	Clinical care cohort/case-control	2012	NA	1.66 (0.95–2.35) per 1000 PY	NA
Hasse et al, 20114	Swiss HIV cohort	Prospective cohort	8444	NA	1.73 (1.26–3.37) per 1000 PY	NA
Rasmussen et al, 201130	Danish HIV cohort	Population-based cohort	5031	45,279	NA	Non-IVDU 1.63 (1.10–2.41) IVDU 1.90 (0.50–7.19)
Chow et al, 201231	Partners HIV cohort	Clinical care cohort	4308	32,423	Total 5.27 per 1000 PY Women 5.02 per 1000 PY Men 5.38 per 1000 PY	Total 1.21 (1.01–1.46) Women 1.76 (1.24–2.52) Men 1.05 (0.84–1.32)
Worm et al, 201293	D:A:D	Prospective cohort	49,737	NA	0.73 (0.62–0.84) per 1000 PY	NA
Vinikoor et al, 201340	UNC HIV clinical cohort	Clinical care cohort	2515	NA	2.26 (1.53–3.21) per 1000 PY	NA
Hemorrhagic strokeb
Justice et al, 2008157	U.S. Veterans Virtual Cohort + California Medicaid dataset	Clinical care cohort	16,541	34,305	0.4 per 1000 PY	2.48 (1.53–4.02)
Durand et al, 201055	RAMQ + Med Echo databases	Population-based cohort	7053	27,681	0.83 (0.58–1.19) per 1000 PY	3.28 (1.75–6.12)
Rasmussen et al, 201130	Danish HIV cohort	Population-based cohort	5031	45,279	N/A	ICH: Non-IVDU 1.47 (0.67-3.20) IVDU 35.62 (4.04-314.17) SAH: Non-IVDU 1.72 (0.80-3.69) IVDU 4.41 (0.81-24.09)
Chow et al, 201356	Partners HIV cohort	Clinical care cohort	4251	35,268	2.29 per 1000 PY	1.85 (1.37–2.47)c

Abbreviations: CI, confidence interval; D:A:D, Data Collection of Adverse Events of Anti-HIV Drugs (D:A:D) cohort; HIV, human immunodeficiency virus; ICD, International Classification of Diseases; ICH, intracerebral hemorrhage; IRR, incidence rate ratio; IVDU, intravenous drug user; NA, not applicable; PY, person-years; RAMQ, Regie de l’assurance-maladie du Quebec; RR, relative rate; SAH, subarachnoid hemorrhage; UNC, University of North Carolina.

^aComposite endpoint of ischemic and hemorrhagic cerebrovascular events.

^bJustice et al and Durand et al studies included both ICH and SAH, while Chow et al study was only ICH.

^cUnadjusted rate ratio; adjusted hazard of ICH among HIV-infected individuals was statistically significantly increased compared with uninfected controls.

Cardiovascular and Ischemic Stroke Risk in HIV-Infected Women

Studies of cardiovascular outcomes in HIV infection have suggested that HIV-infected women may be at relatively greater risk of cardiovascular disease than HIV-infected men.^9,10 In a small study, HIV-infected women had increased noncalcified coronary plaque compared with uninfected women and previously collected data in HIV-infected and uninfected men.³² Virologically suppressed HIV-infected women had urinary levels of eicosanoids consistent with a higher cardiovascular risk profile compared with virologically suppressed HIV-infected men.³³

HIV-infected women may also be at higher risk of stroke. In the Partners HIV cohort, the association between HIV and ischemic stroke was driven largely by women, with an adjusted IRR of 1.76 (95% CI 1.24–2.52) for women compared with 1.05 (95% CI 0.84–1.32) for men.³¹ Intriguingly, in a prospective study conducted only in men from the Multicenter AIDS Cohort Study (MACS), there was no significant difference in incidence of chart-confirmed ischemic strokes compared with uninfected controls over a 15-year period since the introduction of cART.³⁴ Because the relationship between HIV and stroke in other studies is driven by women, the lack of women in the MACS cohort may in part explain the lack of an observed difference in stroke incidence between men with and without HIV infection. In a cohort specifically established to investigate the impact of HIV infection on women in the U.S., the Women’s Interagency HIV Study (WIHS), stroke incidence has not been published. To date, no study has directly explored the underlying causes of this disparity in ischemic heart disease and stroke risk in HIV-infected women; studies on differences in the degree of immune activation by sex,^32,35 independent of viral load, are ongoing. Further epidemiological study of the risk of ischemic heart disease and stroke in HIV-infected women and exploration of potential mechanisms of the risk disparity are a critical area of future study.

Ischemic Stroke Subtypes in HIV–Infected Individuals

The distribution of ischemic stroke subtypes could provide preliminary evidence for the likely pathogenetic mechanisms contributing to increased stroke risk in HIV infection and aid in prediction of recurrent stroke risk as well as prognosis.³⁶ Although subtypes of ischemic stroke have been described in untreated HIV-infected cohorts from earlier in the epidemic,^24,27,28,37 fewer studies postdating the introduction of cART have probed beyond ICD codes or focused on stroke subtype in individuals with treated, well-controlled infection. Ortiz et al reviewed the charts of 82 HIV-infected individuals with strokes (mean CD4 cell count 113 cells/mm³) identified using ICD codes from a major U.S. urban hospital database between 1996 and 2005.³⁸ Among these individuals, 94% (77 of 82) had an ischemic stroke, of which 13% were classified as large artery atherosclerosis, 19% as cardioembolic, 19% as small vessel disease, 23% as strokes due to other determined etiologies including infection and hypercoagulability, and 25% as strokes of undetermined etiology.

Few published studies have focused specifically on treated patients. Of 25 treated HIV-infected individuals with an ischemic event from a clinic in Spain between 1996 and 2008 (mean CD4 cell count 355 cells/mm³), 24% were due to large artery atherosclerosis, 24% to small vessel disease, 8% to cardioembolic disease, 12% to another determined etiology, and 32% to an undetermined etiology.³⁹ In another cohort from North Carolina, the majority of whom were on cART, over 75% of 31 adjudicated ischemic strokes were attributed to large artery atherosclerosis or small vessel disease. Only 10% (3 of 31) of patients, though, were classified as strokes of undetermined etiology, all of whom had insufficient records or incomplete evaluations.⁴⁰ The high frequency of cryptogenic strokes in a majority of these studies,^28,38,39,41 while comparable to the proportion noted in other studies of stroke in the young,^42–44 may be due to practice variation in the work up of stroke etiologies, differential stroke evaluation in patients with HIV, or novel mechanisms for stroke in the context of HIV infection.

Intracerebral and Subarachnoid Hemorrhage in HIV-Infected Individuals

ICH and subarachnoid hemorrhage (SAH) have distinct risk profiles and mechanisms compared with ischemic stroke and cardiovascular disease, yet few studies have specifically evaluated the risk of hemorrhagic stroke in HIV infection. In the pre-cART era, young adults and children with advanced HIV infection were known to develop a cerebral aneurysmal arteriopathy with fusiform dilatation and ectasia of the cerebral vessels, often in the Circle of Willis, with or without SAH.^45–52 Case reports of radiographic stabilization or resolution of aneurysms after initiation of cART^53,54 imply that immunodeficiency may have played a role in the development of these vascular abnormalities.

More recently, Durand et al observed an increased adjusted hazard (HR 3.28, 95% CI 1.75–6.12, p < 0.001) of a composite outcome of ICH and SAH in HIV-infected individuals compared with uninfected age-matched controls using two validated medical databases for all residents of Quebec province between 1985–2007.⁵⁵ The adjusted HR among HIV-infected individuals with an AIDS-defining condition was 7.64 (95% CI 3.78–15.43) compared with 1.99 among those without an AIDS-defining condition. The hazard of ICH/SAH in HIV-infected individuals without AIDS compared with non-infected controls did not reach statistical significance, suggesting that the relationship may have been confounded by conditions associated with advanced immunosuppression, including the aforementioned aneurysmal arteriopathy or opportunistic infections. A comparable increased hazard of ICH among HIV-infected individuals versus controls was seen in the Partners cohort with an IRR of 1.85 (95% CI 1.37–2.47).⁵⁶ Although a protective effect of female sex on the hazard of ICH was observed among HIV-uninfected women, this was lost in HIV-infected women.

In contrast to these two studies, the population-based national Danish study described earlier found the incidence of ICH was higher only among HIV-infected IVDU compared with uninfected controls (adjusted IRR 35.62, 95% CI 4.04–314.17).³⁰ In the U.S. Nationwide Inpatient Sample, the proportion of incident ICH and SAH diagnosed in HIV-infected individuals did not significantly increase between 1997 and 2006 (ICH trend: p = 0.77; SAH trend: p = 0.56).²⁹ Additional research is needed on the epidemiology of hemorrhagic stroke in population-based cohorts with adjudicated outcomes and complete, reliable data on potential confounders of increased ICH risk, such as drug use.

Mechanisms Contributing to Elevated Stroke Risk in HIV Infection

Multiple mechanisms may contribute to the elevated risk of ischemic stroke in HIV-infected individuals, with influence from 1) recognized stroke risk factors, 2) adverse effects of longstanding ART use and 3) HIV-associated inflammation and immune activation.

Traditional Stroke Risk Factors

Several traditional stroke risk factors, including hypertension, dyslipidemia, diabetes mellitus and smoking are more prevalent in HIV-infected individuals, and most are additionally linked to ART use.^57–72 Biomarkers of cardiovascular risk including carotid intima-media thickness (CIMT) may also be abnormal in HIV-infected individuals.^73–75 Multiple observational cohort studies have demonstrated that HIV infection confers 1.5 to 2 times the risk of coronary heart disease and acute myocardial infarction (MI) compared with uninfected controls, independent of recognized cardiovascular risk factors.^6–11 In a recent prospective longitudinal cohort study of over 27,000 HIV-infected U.S. veterans and over 50,000 matched controls, the excess risk of incident MI in HIV infection was 1.48 (HR 95% CI 1.27–1.72) after controlling for cardiovascular risk factors including substance use.¹¹ This excess risk persisted when restricting the analysis to HIV-infected individuals with relatively intact immune function and suppressed viral load, highlighting the potential effect of residual inflammation and immune activation on the risk of cardiovascular disease, even in virally suppressed individuals. Most recent studies of stroke incidence have not observed a statistically significant difference in the prevalence of atrial fibrillation—a strong, independent risk factor for stroke⁷⁶—between HIV-infected compared with uninfected individuals.^30,31 However, a national sample of over 30,000 HIV-infected U.S. veterans found a higher incidence of atrial fibrillation compared with previously published rates from other large epidemiological cohort studies.⁷⁷ Although efforts to modify these risk factors are a key strategy to decrease stroke risk among HIV-infected individuals, studies have consistently demonstrated an excess risk of stroke in HIV infection beyond that explained by traditional risk factors.

Role of Hypercoagulability and Intracranial Vasculopathy in HIV-Associated Stroke

Protein S deficiency has previously been proposed as a mechanism of ischemic stroke in ART-naïve HIV-infected individuals.^24,37 Data are limited supporting its association with arterial stroke even in the general population,⁷⁸ and it has since been discounted as an epiphenomenon of HIV infection unrelated to stroke risk.⁷⁹ Detection of antiphospholipid antibodies has also been reported from two South African cohorts of primarily ART-naïve HIV-infected individuals presenting with stroke^37,80; however, antiphospholipid antibodies are common in HIV infection and are not clearly associated with thrombotic complications.⁸¹ In contrast, elevated tissue-type plasminogen activator antigen and plasminogen activator inhibitor-1 levels have been described in treated, HIV-infected individuals with hyperinsulinemia and fat redistribution (71% of whom were receiving protease inhibitors [PIs]) as compared with controls from the Framingham Offspring Study.⁸² Thus, even in the current era of cART, one possible explanation for excess stroke risk in HIV-infected individuals is impaired fibrinolysis and a resulting hypercoagulable state.

In addition, an intracranial vasculopathy has been described in untreated, severely immunocompromised HIV-infected individuals with and without strokes, often in combination with aneurysmal arteriopathy and, at times, with concomitant infections including varicella zoster virus and cytomegalovirus.^83,84 Pathological examination of affected vessels reveals subintimal fibrosis, disruption of the internal elastic lamina⁸³ and thinning of the media layer even in individuals without clinical or pathological evidence of stroke.⁸⁵ Marked clinical and radiographic improvement of vasculopathy documented after initiation of cART⁸⁶ suggests that this mechanism is likely of considerably less clinical relevance for virologically suppressed individuals on cART.

Antiretroviral Therapy, Immunodepression, and Cerebrovascular Risk

The role of cART in the pathogenesis of stroke in HIV-infected individuals is unclear. Highly effective cART has transformed HIV infection into a chronic disease. Concerns remain, however, regarding the potential adverse health effects associated with use of cART, which is tied to insulin resistance, dyslipidemia, and fat redistribution, and may be an independent risk factor for cardiovascular disease and stroke in HIV-infected individuals.

Numerous studies have explored the link between cART and increased risk of MI,^12,87–91 although two of the largest observational studies have provided conflicting results.^87,89,90 In the Data Collection of Adverse Events of Anti-HIV Drugs (D: A:D) cohort composed of over 30,000 prospectively followed HIV-infected individuals in Europe, the United States and Australia, cumulative exposure to several antiretrovirals (abacavir and two PIs, indinavir and lopinavir/ritonavir) was associated with an increased risk of MI beyond that conferred by dyslipidemia and other metabolic and vascular abnormalities linked to certain medications.^89,90 Recent exposure to abacavir and didanosine were also associated with increased risk of MI. Similarly, in a study by Choi et al of 10,931 HIV-infected U.S. veterans from the Veterans Health Administration Clinical Case Registry (1997–2007), recent abacavir use was associated with an increased hazard (HR 1.48, 95% CI 1.08–2.04) of a composite outcome of cardiovascular events.⁹²

In contrast, from an earlier time period of the Veterans Health Administration Clinical Case Registry (1996–2004), neither cumulative nor recent abacavir use were predictive of MI among nearly 20,000 HIV-infected veterans on cART, although vascular outcomes were not validated as they were in the D:A:D cohort.⁸⁷ One possible explanation for the elevated cardiovascular risk associated with abacavir observed in other studies is confounding by indication. Abacavir may be used more frequently than tenofovir in patients with renal failure, and renal failure is an independent predictor of cardiovascular disease, which could lead to a spurious association between abacavir and increased cardiovascular risk. Indeed, after controlling for renal disease, abacavir use and tenofovir use, when compared with the reference group of a regimen containing neither abacavir nor tenofovir or no ART use at all, actually protected against MI.

Fewer data are available assessing the contribution of cART use to stroke risk, and findings are inconsistent. An association between any cART use and a composite outcome of coronary heart disease and stroke in the D:A:D cohort raised the specter of elevated cerebrovascular risk due to ART, with a 26% higher adjusted risk of a vascular event per year of ART exposure.¹⁴ However, when vascular outcomes were evaluated individually in the D:A:D cohort, no association was identified between cumulative or recent exposure to either abacavir or didanosine and stroke, despite their association with an increased risk of MI.⁹⁰ Power was substantially lower though, for the stroke analysis due to fewer strokes (196 events) compared with cardiovascular events (693 events). In contrast, in the previously cited Choi et al study, abacavir use was associated with more than twice the hazard of CVD (HR 2.10, 95% CI 1.20–3.66).⁹² Similarly, in the population-based national Danish study, abacavir use among HIV-infected non-IVDU was the only antiretroviral medication associated with a higher risk of cerebrovascular events (adjusted HR 1.66, 95% CI 1.03–2.68), though confounding by indication may have inflated the apparent association.

Other studies have found an association between cART use and a reduction in stroke risk. One of the earliest studies evaluating the impact of the introduction of cART performed in a national Veterans Affairs database found a reduction in admissions for cardiovascular disease and CVD from 1.7 per 100 PY in 1995 to 0.9 per 100 PY in 2001 and a comparable decline in overall death rate in the HIV-infected population from 21.3 per 100 PY in 1995 to 5 per 100 PY in 2001.¹² Neither rates of admissions nor of deaths related to cardiovascular disease or CVD rose with increasing exposure to cART. In a case-control study of HIV-infected individuals on cART with a first-ever ischemic stroke compared with those without a stroke, longer duration of cART was associated with 3% lower odds of cerebrovascular events for every additional month of cART exposure (OR 0.97, 95% CI 0.96–0.99).³⁹ Likewise, in the Partners HIV cohort, longer duration of any ART use was associated with a 21% reduction in risk of ischemic stroke (HR 0.79, 95% CI 0.71–0.88).³¹

ART use in these cases may function as a proxy for control of HIV infection, with achievement of viral suppression and restoration of immune function conferring decreased stroke risk. Cumulative exposure to cART, however, may lead to a rise in vascular risk over time, the effect of which will not be detectable for years to come. This hypothesis is supported by recent results from the D:A:D cohort which did not demonstrate a statistically significant association between current or recent PI exposure and a composite outcome of sudden death and first ischemic stroke; however, cumulative exposure to PIs increased the risk of the composite outcome (adjusted risk ratio of 1.06 per year of exposure, 95% CI 1.01–1.11).⁹³

Consistent with this hypothesis, the degree of immunosuppression appears associated with biomarkers of cardiovascular risk (i.e., subclinical carotid atherosclerosis, endothelial dysfunction) and incident cardiovascular events.^73,94–99 In the Partners cohort study, higher HIV viral load was associated with a higher rate of ischemic stroke, and a greater CD4+ T-cell count exerted a protective effect against ICH.^31,56 In the national Danish study, in HIV-infected individuals without a history of IVDU who had not initiated cART, a CD4+ T-cell count ≤ 200 cells/mm³ was associated with more than twice the risk of a cerebrovascular event as compared with higher T-cell counts.³⁰ Similarly, both viral load and CD4+ T-cell count were associated with an increased rate of cerebrovascular events in a North Carolina cohort.⁴⁰ In an adjusted model, HIV-infected individuals with a recent detectable viral load had more than 3.5 times the rate of cerebrovascular events compared with those with an undetectable viral load, while those with a recent CD4+ T-cell count < 200 cells/mm³ had over 2.5 times the rate of events compared with those with higher CD4+ T-cell counts.

These findings reinforce the hypothesis that viral load and CD4+ T-cell count—markers of control of HIV infection—may predict stroke risk in the HIV-infected population. Alternatively, cART may lead to a decline in the incidence of strokes by reducing misclassification bias related to stroke-like events and/or stroke mimics typically seen in profound immunodeficiency. In the D:A:D cohort, investigators observed that doubling the most recent CD4+ T-cell count was associated with a lower adjusted RR of stroke (RR 0.75, 95% CI 0.70–0.81).¹³ The observed association, however, was less strong, though still significant, after removing stroke-like events and stroke mimics from the analysis (RR 0.81 for each doubling of CD4 cell count, 95% CI 0.74–0.89).

Considerable controversy remains around vascular risk and cART, and in particular between abacavir and risk of cardiovascular disease and stroke.^100,101 Although particular attention has been paid to abacavir, use of other antiretrovirals has not demonstrated an increased risk of stroke,¹⁰² though few of the newer agents have been studied.

Effect of Inflammation and Immune Activation on Stroke Risk

Inflammation, involved in the pathogenesis of atherosclerosis in the general population,^103,104 may also be implicated in stroke pathogenesis in HIV infection. The hypothesis that chronic infections may play a role in the development of inflammation and vascular outcomes including stroke is not new. Elkind et al devised a weighted index of serology titers for five common pathogens, including herpes simplex 1 and 2, associated with an elevated adjusted risk of stroke and with carotid plaque thickness in a community-based prospective cohort study.^105,106 Likewise, HIV-associated inflammation and immune activation have been implicated in the observed excess cardiovascular risk in HIV infection. The landmark Strategies for Management of Antiretroviral Therapy (SMART) trial, designed originally to evaluate episodic cART guided by CD4+ T-cell count as a strategy to reduce adverse effects associated with ART, demonstrated the important and unexpected finding that HIV-infected individuals receiving episodic cART had higher rates of cardiovascular events.¹⁰⁷

Inflammatory and coagulation markers were elevated in those on episodic therapy and were associated with all-cause mortality¹⁰⁸ as well as future cardiovascular events including stroke.¹⁰⁹ Other studies have shown a similar association between markers of inflammation and immune activation and cardiovascular biomarkers and outcomes in HIV infection.^110–114 Interestingly, even in treated individuals with well-controlled infection, evidence of inflammation and immune activation are still present.^112,114,115 Whether this level of inflammation and immune activation results in ongoing, accelerated vascular aging and impacts cerebrovascular risk in HIV-infected individuals despite virologic suppression remains to be seen.

Vascular Risk and HIV-Associated Cognitive Impairment

The relationship between vascular risk factors, atherosclerosis and cognitive impairment has gained increasing attention in the general population^116,117 and suggests shared genetic, environmental, and vascular mechanisms of dementia. A parallel association has been explored in HIV-associated cognitive impairment, which remains highly prevalent in the era of cART^118–120 and is associated with reduced quality of life and functional independence.¹²¹

Valcour et al found that diabetes, defined either by self-report or standard fasting and/or random glucose levels, increased the adjusted odds of HIV-associated dementia more than fivefold in the Hawaii Aging with HIV cohort.¹²² In the same cohort, insulin resistance also predicted worse cognitive impairment, particularly in older individuals.¹²³ In contrast, in the WIHS study of HIV-infected women and uninfected controls, severity of insulin resistance among relatively young HIV-infected women was not associated with most neuropsychological tests performed.¹²⁴ In the neurology substudy of the SMART trial, a history of cardiovascular disease resulted in a sixfold higher odds of baseline neurocognitive impairment in individuals with well-controlled HIV infection (median CD4 cell count 538 cells/mm³).¹²⁵ Other vascular risk factors (e.g., use of an antihypertensive agent, higher total cholesterol) also predicted worse neuropsychological test scores, whereas markers of control of HIV infection were not associated with test performance.

In addition to vascular risk factors, the association between biomarkers of cardiovascular risk, such as CIMT, and cognitive impairment in HIV infection have been examined, although few studies have provided convincing evidence of a strong association.^126–129 In a cross-sectional analysis of 635 HIV-infected and uninfected participants from the MACS cardiovascular and cerebrovascular substudy, CIMT or coronary artery calcium were associated with impaired psychomotor speed or memory in adjusted models; however, HIV status itself had no significant impact on cognitive function.¹²⁷ In the relatively young WIHS cohort, CIMT and presence of carotid lesions were predictive of worse performance on the Stroop interference task, a measure of executive function, but HIV status was not.¹²⁸ The reverse was true for performance on the symbol-digit test, a measure of information processing speed—HIV infection predicted worse performance but no association between subclinical carotid atherosclerosis and test performance was observed. In contrast, in 245 consecutively enrolled HIV-infected outpatients from two clinical centers, diabetes and higher CIMT were independently associated with lower global cognitive performance,¹²⁹ even when restricting the analysis to virologically suppressed individuals. Lastly, age-related chronic small vessel ischemic changes, which are strongly tied to stroke and dementia in the general population,^130–132 have also been described in HIV-infected individuals,¹³³ but the link, if any, to HIV-associated cognitive impairment in the current era of cART is unclear. One recent study found a correlation between the volume of white matter hyperintensities and executive function in a cohort of older HIV-infected individuals.¹³⁴ White matter hyperintensity volume was not associated with degree of immunosuppression or duration of HIV infection. Smokers and hypertensive participants, however, had a greater burden of white matter disease, and HIV infection appeared to modulate these effects.

Tools that directly assess cerebrovascular function may provide additional information on the potential interaction between vascular risk factors, cerebrovascular injury, and cognitive impairment in HIV infection. Cerebral vasoreactivity, a measure of intracranial endothelial function¹³⁵ that can be assessed using transcranial Doppler, has been shown to be diminished in HIV-infected individuals compared with uninfected controls.¹³⁶ Future studies evaluating dynamic cerebrovascular function should focus on the association between cerebral vasoreactivity, neurovascular coupling, and both stroke and cognitive impairment. This will complement current neuroimaging research on cerebral blood flow and metabolism, which has been performed primarily in the context of HIV-associated cognitive impairment.¹³⁷ In summary, while intriguing associations between vascular risk and cognitive impairment in HIV infection have been detected, a strong link between the two has not been established. Prospective, clinical-pathologic-radiologic studies are required to understand the complex interplay between vascular and HIV-related risk factors in the development of cognitive impairment in this population.

Vascular Risk Prediction and Reduction in HIV Infection

Stroke Outcomes

Exceedingly limited data are available on stroke outcomes in HIV infection in the current era of cART. Vinikoor et al reviewed the charts of all HIV-infected individuals in a clinical care cohort in North Carolina with an identified cerebrovascular event. Of 53 confirmed events, two patients (4%) died within 28-days and 15% at one-year, although mortality rates from an uninfected comparator group were not available.⁴⁰ No information regarding functional outcome after stroke was reported. In a prospective study in Blantyre, Malawi of 147 first-ever strokes over a 14-month period, of whom 50 occurred in HIV-infected individuals, neither mortality nor functional outcome at one-year differed by HIV status.¹³⁸

Risk Prediction

The accuracy of traditional risk prediction algorithms in determining cardiovascular risk in an HIV-infected population has been a topic of considerable interest.^139,140 The Framingham Risk Score (FRS), one of the most widely used cardiovascular risk assessment tools, has been applied to the D:A:D cohort to predict risk of MI compared with observed events.¹⁴¹ Key variables, including left ventricular hypertrophy (LVH), smoking, systolic blood pressure, and cholesterol were either unavailable or partially missing for the cohort, and were assumed to be absent or were imputed in participants. Predicted risk was higher than observed in cART naïve individuals, and lower than observed in participants who had received cART, although the 95% CI of the predicted rates overlapped with observed rates. A cardiovascular risk assessment tool specific for HIV-infected individuals, which included stroke as part of a composite outcome, has been developed in the D:A:D cohort.¹⁴² In addition to traditional vascular risk factors, the model also included exposure to two PIs and abacavir; neither CD4 cell count nor HIV viral load were included in the prediction model due to lack of significant associations with the outcomes. When compared with the FRS, the HIV-specific prediction tool was better calibrated (i.e., ability to accurately predict risk of cardiovascular events), with a mean ratio of predicted to observed events of 0.95 (range 0.76–1.42). The generalizability of the risk prediction tool to other HIV-infected populations, however, is unknown as the tool has only been validated in a subset of the D:A:D cohort.

Mateen et al evaluated the utility of a stroke-specific FRS developed in the general population¹⁴³ in the MACS cohort.¹⁴⁴ Strokes in HIV-infected and uninfected MACS participants were ascertained by self-report, review of causes of death, or when reviewing medical records for other purposes. The stroke-specific FRS was then calculated 10 years before the stroke event. Despite a higher incidence of stroke in the HIV-infected participants (3.3 vs. 1.7 per 1000 PY), the mean 10-year stroke-specific FRS was lower for HIV-infected participants (4.9 vs. 6.6%). Importantly, inclusion of unconfirmed self-reported strokes may have led to an overestimate of strokes in the cohort, particularly as the authors had previously reported no difference in the incidence of chart-confirmed strokes in the MACS cohort by HIV status,³⁴ although this assumes stroke misclassification was more likely to occur in HIV-infected compared with uninfected participants. Furthermore, data on both atrial fibrillation and LVH were unavailable in the cohort and counted as absent, as were any additional missing risk factors. There remains a pressing need to develop a stroke risk prediction tool tailored to HIV-infected individuals to optimize future preventive strategies for this at-risk group.

Strategies for Stroke Prevention

Without specific data guiding primary and secondary stroke prevention or acute stroke therapy in HIV infection, the approach to prevention and management of stroke in HIV-infected individuals should follow current national guidelines. Recognized stroke risk factors, including hypertension, dyslipidemia, and smoking, are prevalent in HIV infection and should be managed aggressively in collaboration with HIV providers and pharmacists who are knowledgeable about potential drug interactions. In fact, aspirin and lipid-lowering therapy may be underutilized in HIV-infected individuals.^145–148 Moreover, the effect of these therapies may differ in HIV-infected individuals compared with the general population. Regardless of cART regimen, HIV-infected individuals had smaller reductions in low-density lipoprotein (LDL) cholesterol levels after initiation of a statin compared with uninfected controls.¹⁴⁹ Similarly, the utility of aspirin may extend beyond its antithrombotic properties in HIV infection as it has been shown to reduce markers of T-cell and monocyte activation in virologically suppressed HIV-infected individuals when compared with uninfected controls.¹⁵⁰ Lastly, a systematic and comprehensive approach to stroke care in resource-limited settings is urgently needed.¹⁵¹

Evaluating strategies to address risk factors unique to HIV-infected individuals, including persistent inflammation and immune activation despite cART and virologic suppression, may also be essential to stroke prevention in the future, particularly as individuals live longer with well-controlled infection. Statin use, for example, has been shown in a randomized controlled trial (RCT) to reduce cardiovascular event rates and all-cause mortality among apparently healthy men and women with LDL less than 130 mg/dL and high sensitivity C-reactive protein greater than or equal to 2.0 mg/L.¹⁵² Although no RCT of the effect of statin use on cardiovascular events has been performed in an HIV-infected population, observational studies suggest it may also reduce all-cause mortality in HIV-infected individuals,^153,154 although the benefit in those without cardiovascular comorbidities may be limited.¹⁵³ Other strategies under investigation include timing of initiation of cART, intensification of cART regimens, and use of anti-inflammatory agents and methods to reduce microbial translocation in the gut.^155,156

Conclusion

The shifting landscape of the care of HIV-infected individuals requires clinicians to be knowledgeable about chronic, non-infectious complications of HIV infection, including vascular disease and stroke. Up-to-date research on the epidemiology and mechanisms underlying these chronic complications in HIV infection is essential for the optimization of stroke prevention and treatment strategies for the future and for the development of management guidelines for widespread dissemination to successfully meet the challenge of caring for CVD in HIV-infected individuals worldwide.